Feature: Support LJA for genome size estimation in helper command

src/helper.rs

@@ -37,12 +37,18 @@ use crate::subsample::parse_genome_size;
 #[allow(clippy::too_many_arguments)]
 pub fn helper(task: Task, reads: PathBuf, out_prefix: Option<PathBuf>, genome_size: Option<String>,
               threads: usize, dir: Option<PathBuf>, read_type: ReadType,
-              min_depth_abs: Option<f64>, min_depth_rel: Option<f64>, extra_args: Vec<String>) {
+              min_depth_abs: Option<f64>, min_depth_rel: Option<f64>, extra_args: Vec<String>,
+              assembler: Option<String>) {
     check_if_file_exists(&reads);
     let (dir, _guard) = get_working_dir(dir);
 
     if task == Task::GenomeSize {
-        genome_size_raven(reads, threads, dir, extra_args);
+        let assembler = assembler.unwrap_or_else(|| "raven".to_string());
+        match assembler.to_lowercase().as_str() {
+            "raven" => genome_size_raven(reads, threads, dir, extra_args),
+            "lja" => genome_size_lja(reads, threads, dir, extra_args),
+            _ => quit_with_error(&format!("unknown assembler: {assembler}")),
+        }
         return;
     }
 
@@ -403,6 +409,28 @@ fn genome_size_raven(reads: PathBuf, threads: usize, dir: PathBuf, extra_args: V
 }
 
 
+fn genome_size_lja(reads: PathBuf, threads: usize, dir: PathBuf, extra_args: Vec<String>) {
+    check_requirements(&["lja"]);
+
+    let assembly_path = dir.join("assembly.fasta");
+
+    let mut cmd = Command::new("lja");
+
+    cmd.arg("--threads").arg(threads.to_string())
+       .arg("--output-dir").arg(&dir)
+       .arg("--reads").arg(&reads);    
+
+    for token in extra_args { cmd.arg(token); }
+
+    run_command(&mut cmd);
+
+    if is_fasta_empty(&assembly_path) {
+        quit_with_error("LJA assembly failed");
+    }
+    println!("{}", total_fasta_length(&assembly_path));
+}
+
+
 fn redbean(reads: PathBuf, out_prefix: &Path, genome_size: Option<String>, threads: usize,
            dir: PathBuf, read_type: ReadType, extra_args: Vec<String>) {
     // https://github.com/ruanjue/wtdbg2

src/main.rs

@@ -233,6 +233,10 @@ enum Commands {
         #[clap(long = "args", value_parser = clap::builder::NonEmptyStringValueParser::new(),
                num_args = 1.., action = clap::ArgAction::Append, allow_hyphen_values = true)]
         args: Vec<String>,
+
+        /// Assembler to use (for genome_size task)
+        #[clap(long = "assembler")]
+        assembler: Option<String>,
     },
 
     /// resolve repeats in the the unitig graph
@@ -349,9 +353,9 @@ fn main() {
             gfa2fasta::gfa2fasta(in_gfa, out_fasta);
         },
         Some(Commands::Helper { task, reads, out_prefix, genome_size, threads, dir, read_type,
-                                min_depth_abs, min_depth_rel, args }) => {
+                                min_depth_abs, min_depth_rel, args, assembler }) => {
             helper::helper(task, reads, out_prefix, genome_size, threads, dir, read_type,
-                           min_depth_abs, min_depth_rel, args);
+                           min_depth_abs, min_depth_rel, args, assembler);
         },
         Some(Commands::Resolve { cluster_dir, verbose }) => {
             resolve::resolve(cluster_dir, verbose);